Investigation of the application of Statistical Process Control into Low Volume Manufacturing

Abstract

Statistical process control (SPC) into Low Volume Manufacturing environment face a challenge applying SPC techniques. SPC is commonly used for quality control and improvement in the manufacturing sector. In the early 1920s, Dr Walter Shewhart developed the control chart employed to monitor a process over time, where the first data is collected and then plotted on a graph. Moreover, a control chart is composed of a Central Line (CL), the Upper Control Limit (UCL) and the Lower Control Limit (LCL). Parameters and control limits are calculated to analyze the control chart, requiring twenty to twenty-five subgroups of data, with three to five values per subgroup, or at least sixty measurements. However, collect this amount of data is difficult in certain production processes, where the lot size could even be one and it could take weeks or months to accumulate enough data to estimate the process parameters.
Statistical process control is a challenge in some scenarios such as startup production, different or individual parts in the same production line, or production of customized products. In these cases, there is not enough amount of data to compute the parameters to monitor the process. Therefore, special techniques and statistical methods are required. Some authors developed self-starting control charts and alternative methods for short-run production, e.g. Q charts, Exponentially Weighted Moving Average (EWMA) and Cumulative sum (CUSUM). This thesis studies the performance of these SPC tools, implementing a Low Volume Statistical Process
Control (LV-SPC) model through an Excel spreadsheet, analyzing the production process data from companies that are performing low volume manufacturing. This work provides an interpretation and explanation about statistical process
control into low volume manufacturing, analyzing the application of different SPC methods developed for short production runs based on data collected from different companies. Data collected was processed to individual measurements from the process deviation rather than the mean values. Converting the data to individual values the SPC methods for low volume manufacturing are viable to use. Also, performance, effectiveness, and how it can be further implemented were discussed.